There are a lot of questions being asked about 6G, despite the fact that (or perhaps because) it’s likely half-a-decade away. One of the questions is how it should treat the RAN, whether “open RAN” should be mandated in 6G. Telcos I chat with favor that position by 2:1, but interestingly even those who do tell me that they don’t believe that it will really change the RAN or vendor landscape much. A decent number even think that the debate is destructive to addressing other, more important-to-telco, questions.
What vendors think about this is more complicated, as a Light Reading story indicates. I don’t have any credible contacts with Huawei or ZTE at this point, but some telcos who do tell me the same story the article cites, and that to me suggests some truths. Truth One: Incumbent vendors believe they’d be better off without open RAN. Truth Two: Most incumbent vendors realize that the initiative failed to have a major impact in 5G, and will almost surely do the same for 6G. Truth Three: “almost surely” isn’t necessarily enough of a guarantee.
Huawei is somewhat unique among the big mobile infrastructure players in that it faces a political pressure to drop its gear from networks, and it has a larger share of emerging-market telcos than it rivals. This combination means that there is more pressure on their incumbencies, and a greater chance that the telcos under the pressure would look at an open strategy to save money.
But let’s face it, we live in a world dominated by clicks, and nobody really wants to stand up in public against openness. That doesn’t mean they’d give it more than lip service, though, and that in turn means that it’s fair to ask whether the whole idea has, in the telco world, any value at all. Let’s look at that question first.
Telcos, enterprises, and pretty much everyone who consumes technology really want a system in which everything is plug and play—no integration, no complexity of operations. They also want the best possible technology at the lowest possible price, preferably free. Sadly, we live in a world of unsatisfied wants as much as we live in one dominated by clicks. The fact is that the benefits of openness demand multiplicity of players and competition among approaches, and if you forswear the notions of integration and resolving operational complexity, you won’t have that. A bunch of open players emerges when you can have best-of-breed competition at the micro-component level, and many/most of the competitors then would likely be micro-competitors who had to be combined to provide a complete solution.
How do you eliminate integration and operational complexity in an open infrastructure? The only possible solution is AI, which to me means that pushing open 6G means betting that by the time it comes along, autonomous AI operation of infrastructure will be accepted broadly, and that the open strategy would be designed to maximize the chances of that happening. We do not know whether the first is true, and we should know that at present we have no credible idea of how that AI optimization could evolve.
That last point is important, because if we had an open RAN initiative in/for 6G that could address the point, it might justify the time spent on promoting the concept and the divisive and distracting impact of its inclusion. Given my experience with standards bodies and industry groups, I think a formalized inclusion of open RAN in 6G would virtually exclude all chance of progress in the AI optimization space. I also think that involving standards/industry groups is likely to do nothing but slow and dilute the effort, which is why I think the AMD/GSMA initiative is almost certain to do little or nothing to help.
This is something that Nvidia should undertake, instead of issuing platitude studies, and contribute their results in an open release. They’re promising an open-source 6G model, but that’s not enough to resolve the integration/operations concerns. An AI operations interface model could influence development of AI tools, and their inclusion in 6G deployments.
Why not the 6G vendors? It goes back to the truth that nobody who’s an incumbent has anything to gain from an open RAN adoption, nor to any AI initiatives that would facilitate multi-vendor deployment. The vendors would surely accept AI tools in a public sense (as they did with Open RAN) if someone like Nvidia pursued the strategy, but if they did something on their own they’d justifiably focus it on their own product line, and gain from their investment.
So, OK, there could be a value in an open element to 6G, but IMHO it would disappear if it were to be confined to the RAN. There’s no such thing as a partially open infrastructure, limited integration, limited operations complexity, as far as telcos are concerned. The benefits of openness would dissipate or even disappear if they were limited to a piece of infrastructure. You can see that attitude in how Open RAN for 5G has failed to deliver; telcos buy infrastructure not RAN infrastructure, and so you need to extend the thing you expect to facilitate openness to the entire mobile infrastructure. Which, of course, makes it a lot more complicated.
My conclusion here is that 6G should not mandate open RAN, or open anything, at this point. Absent a way to totally address the integration/operations issues of telcos, such a decision would not have any better outcome than Open RAN has had in 5G. It would, again IMHO, likely have a worse outcome, because fighting 5G-like battles in 6G risks having the whole initiative fail for the same reasons that 5G did—which we can summarize as being a supply-side-field-of-dreams mindset.
Future mobile services can’t be profitable if they don’t offer differentiation based on QoS, meaning that they support applications that require a level of QoS better than the best-efforts mindset of today’s mobile broadband. That doesn’t mean that these future services can skate by simply offering the QoS; they have to be able to promote the applications. That, in turn, means they have to feed an active project process or launch one. Since the latter is hardly possible for a telco, they have to hope for the former.
My discussion with enterprises on the evolutionary-project-centricity of their technology spending told me something new, which was that if you can’t launch a credible project to change tech adoption you could only feed one that got started without you. They dismiss the notion of revolutions because they’d require revolutionary spending, revolutionary project approval cycles, and impose revolutionary risk. That, I believe, means dealing with the evolution of IoT applications in process management and control from being premises-based to something more metro-scale. If I’m correctly analyzing what enterprises are telling me, that evolution is the only path to 6G success.
